# Variations in Canopy Cover and Its Relationship with Canopy Water and Temperature in the Miombo Woodland Based on Satellite Data

^{1}

^{2}

^{3}

^{4}

^{*}

## Abstract

**:**

^{2}) ranging between 0.67 and 0.96. However, the coefficients of estimates showed the canopy water content (i.e., NDII) to have had the largest percentage of the interactive effect on the variations in canopy cover regardless of the proxy used i.e., LAI or NDVI. From 2009–2018, the NDII (proxy for canopy water content) showed no significant (at alpha level 0.05) trend. However, there was a significant upward trend in LST (proxy for canopy temperature) with a magnitude of 0.17 °C/year. Yet, the upward trend in LST did not result in significant (at alpha level 0.05) downward changes in canopy cover (i.e., proxied by LAI and NDVI). This result augments the observed least determinant factor characterization of temperature (i.e., LST) on the variations in canopy cover as compared to the vegetation water content (i.e., NDII).

## 1. Introduction

^{2}[1,2]. The woodland forms the transition zone between the tropical rainforest and the African Savannah. Being a transition zone, it is sensitive to climate change where dry-out could possibly trigger ecosystem shifts. The word Miombo is used to describe Brachystegia, one of the many species found across the ecosystem. The Miombo ecoregion plays a crucial role in the food, water, and energy nexus in sub-Sahara Africa and maintain carbon stocks, thus regulating climate [1,3,4,5]. Furthermore, studies [6,7] have shown that forests, such as the Miombo Woodland, plays a critical role in the water balance by affecting the land surface water interaction such as precipitation canopy interception and extraction of soil water via the transpiration process. However, rising temperatures, changing precipitation regimes, and changes in the amount of carbon dioxide are expected to affect phenology, composition, structure, distribution, and succession processes of forests [1,8]. Thus, the relationships of canopy cover with variables such as canopy water and temperature must be well understood and consequently taken into account in climate and hydrological modelling [9] in the Miombo Woodland.

^{2}0.96) in NDVI values followed by rainfall (R

^{2}= 0.35). Further, Chidumayo [14] observed a negative relationship between NDVI and maximum temperature. Tian et al. [19] studied seasonal variations in ecosystem-scale plant water storage and their relationship with leaf phenology. They found that in the Miombo Woodland transpiration co-varied with LAI seasonal variations. They further observed that plant water storage played a critical role in “buffering seasonal dynamics of water supply and demand, as well as sustaining fresh leaves formed before the rain”.

## 2. Materials

#### 2.1. Study Site

#### 2.2. Description of Data Sets

_{a}) (MOD16A2/MYD16A2) [49] at 8-day interval and 500 m spatial resolution was also obtained using the MODIS global subset tool. The MOD16A2/MYD16A2 uses the Penman–Monteith equation [57,58] and is a sum of night-time and day-time actual evaporation. It is the sum of the evaporation from the wet canopy surface, the transpiration from the dry canopy surface, and the evaporation from the soil surface [49]. In this study, the term evaporation is used instead of evapotranspiration. This is based on the arguments by Miralles et al. [59] and Savenije [60] in which they showed that the term evapotranspiration has limitations and can even be considered as misleading.

_{a}, LAI, and LST data sets. Polygons for each of the Miombo Woodland location(s) (Figure 1) were used to extract the required data sets from January 2009 to December 2018. Raw filtered data provided on the global subset tool was used to estimate the 5th and 95th percentiles for the variables. The focus of the study was the dry season and had most pixels with acceptable quality pass of 100%.

## 3. Methods

#### 3.1. Study Approach

_{a}) and rainfall are closely linked to the LAI, LST, NDII, and NDVI [30,31,32,36,65,66], they were used to augment the observations in the general patterns of the LAI, LST, NDII, and NDVI, but were not included in the statistical analyses as predictor or response variable. Furthermore, the phenology calendar was used to augment the LAI and NDVI observations.

^{2}) area in the Miombo Woodland at Nsansala and Mutinondo conservancy areas (Figure 1). The conservancy areas were selected due to availability of a dense Miombo Woodland which is protected and is with minimal anthropogenic disturbances. The data for each variable from the three polygons was merged into one data set per variable and analyzed. To avoid mixing variables values from different land cover types (i.e., grassland, bare soil, and other forest types) with Miombo Woodland the assessment was restricted to a dense Miombo Woodland (Figure 1). Hence, the assessment was not done at the entire Miombo ecosystem level in Africa but restricted to this dense Miombo woodland in the Luangwa Basin in Zambia.

#### 3.2. Statistical Approaches Used to Analyse the Data

#### 3.2.1. Testing for Outliers with the Dixon Test

#### 3.2.2. Assessing Correlation of Variables with the Pearson R Approach

#### 3.2.3. Assessing Variations in the Variables with the RM-ANOVA

#### 3.2.4. Correlated Component Regression Linear Model (CCR.LM)

_{1}, X

_{2},……, X

_{P}. The CCR-linear algorithm is a generalization of the CCR-logistic and is generally performed in three steps. Step 1: the first component S

_{1}captures the effects of prime predictors that have direct effects on the outcome. It is an average (ensemble) of all 1-predictor effects: Step 2: the second component S

_{2}is correlated with S

_{1}and captures the effects of suppressor variables (proxy predictors) that improve prediction by removing extraneous variation from S

_{1}. Step 3 estimates the 2-component model using S

_{1}and S

_{2}as predictors. The predictions (coefficients of estimates) for Y in the K-component CCR model are obtained from the ordinary least square regression of Y on ${S}_{1}$,……,${S}_{k}$. The regression coefficient for predicting a variable is simply the weighted sum of loadings, where the weights are the regression coefficients for the components in the K-component model. The step-down variable reduction step for a given K-component model eliminates the least important predictor variable, where importance is quantified as the absolute value of the variable’s standardized coefficient. If a predictor variable is found to have the least absolute value of the standardized coefficient, the predictor is excluded and the steps of the CCR estimation algorithm repeated on the reduced set of predictors. The regression correlation coefficient of determination (R

^{2}) and the normalized mean square error (NMSE) [50,70] are used to assess the performance of the CCR.LM in predicting the response variable(s) with a given combination of predictor variables.

#### 3.2.5. Testing for Homogeneity with Pettit Test

#### 3.2.6. The Mann–Kendal Trends Test and the Sens’ Method

## 4. Results

#### 4.1. Assessment of Outliers with the Dixon Test and Z-Score Approach

_{Max}values to be outliers. The LAI

_{Max}outliers were lowest values of 1.1 observed on the 7th and 20th of July 2009, 20th and 13th of August 2011, and also on the 13th and 19th July 2012. These observations can possibly be attributed to underestimation by the LAI algorithm as the observations were for shorter periods between two points with higher values on both ends (before and after). Furthermore, these outliers in LAI

_{Max}were not linked to any changes in the LST, NDII, or NDVI on the same dates or period. For the LST, some outliers were mostly flagged in the rainy season, which is possibly due to the associated challenges in assessing LST in hot-humid environments. The rainy season LST “outliers” could also indicate drought periods during the rainy season though this was discounted based on the counterchecking with rainfall, LAI, NDII, and NDVI values for the same dates/period(s). In general, the test for outliers showed that there were no extreme events in the variables examined for the period 2009–2018 in the study area.

#### 4.2. Pattern Observed in Phenology, Canopy Cover, Water Content and Temperature

_{a}were observed at the end of the rainy season in April.

_{a}and the canopy cover (i.e., LAI and NDVI) and the canopy water content (i.e., NDII) across seasons (Figure 3 and Figure 4).

#### 4.3. Analysis of the Observed Patterns with the RM-ANOVA Test

#### 4.4. Pearsonr Correlation Analysis of the Relationship of Variables

_{Mean}and NDII

_{Mean}, as well as between LAI

_{Mean}and LST

_{Mean,}indicating that changes in canopy cover (i.e., LAI

_{Mean}) lagged behind changes in canopy water content (i.e., NDII

_{Mean}), as well as changes in canopy temperature (i.e., LST

_{Mean}). The changes in the canopy cover was not only affected by the canopy water (i.e., NDII

_{Mean}) and temperature (i.e., LST

_{Mean}) in a given day but by cumulative canopy water and temperature conditions. Generally, canopy cover (i.e., NDVI

_{Mean}) showed synchronism with canopy water content (i.e., NDII

_{Mean}) throughout the year. Canopy cover (in the case of LAI

_{Mean}) only showed strong correlation with canopy temperature during the hot-dry season. Statistically, at the annual scale, the canopy water content (i.e., NDII

_{Mean}) relationship with canopy cover (i.e., LAI

_{Mean}and NDVI

_{Mean}) was strongly positive with r = 0.78, 0.97 with p-value < 0.0001, respectively (Figure 5 and Table 3).

_{Mean}) versus canopy cover (i.e., LAI

_{Mean}and NDVI

_{Mean}) correlation was strongly positive except for the NDII-LAI relationship, which showed a low coefficient of correlation (i.e., r = 0.29 and p-value < 0.008) for the rainy season (Tables S4–S7 in the supplementary data). However, during the hot-dry season, strong NDII-LAI correlation was observed with r = 0.70 and p-value < 0.0001 (Table S7). Generally, NDII-NDVI correlations were stronger (i.e., r > 0.75) across all seasonal scales (Tables S4–S7 in the supplementary data). Table 3 and Tables S4–S7 in the supplementary data gives statistics of the Pearson r correlation analysis of the canopy temperature (i.e., LST) correlation with canopy cover (i.e., LAI and NDVI). At the annual scale, the canopy cover (i.e., LAI

_{Mean}as a proxy) negatively covaried with canopy temperature with r = −0.11 and p-value < 0.83. When canopy cover was proxied by NDVI a negative covariation with canopy temperature of r = −0.41 and p-value < 0.0001 was observed. During the rainy season, the LST-NDVI relationship showed a correlation coefficient of r = −0.44; and p-value < 0.0001, while LST-LAI correlation had r = −0.52; and p-value < 0.001.During the dry season (May–October), the LAI-LST correlation was positive but weak and statistically insignificant with r = 0.04 and p-value = 0.56. The NDVI-LST correlation was negative but relatively stronger with r = −0.68 and p-value < 0.0001. During the cool-dry season (May–July) the NDVI-LST correlation was positive but relatively weak (r = 0.26, p-value < 0.0046), and was the same as the LAI-LST relationship (r = 0.26, p-value < 0.0044) (Tables S4–S7 in the supplementary). However, the hot-dry season (August–October) showed stronger positive LAI-LST correlation (r = 0.62, p-value < 0.0001), while the NDVI-LST relationship was negative (r = −0.25, p-value = 0.0097) (Tables S4–S7 in the supplementary). Generally, at the annual scale, LST

_{Min}values showed relatively stronger correlation with LAI

_{Min}and NDVI

_{Min}compared to the LST

_{Max}and LST

_{Mean}values (Table 3).

#### 4.5. Regression of Canopy Cover Relationship with Canopy Water Content and Canopy Temperature Using the CCR.LM

^{2}) were used to assess the estimate of the overall deviations between predicted and “observed” values, thus ascertaining the performance of the model. The unstandardized coefficients ($\widehat{\xdf}$) were used to analyze which of the input variables (i.e., LST or NDII), at each run, had a significant effect on each response variable (i.e., LAI and NDVI) at different time scales, i.e., annual and dry season

^{2}= 0.61) of the variations in LAI and 93% (R

^{2}= 0.93) in NDVI. LST as a single variable accounted for 13% in LAI variations and 17% in NDVI variations (Table 4). A combination of the LST with NDII accounted for about 67% (R

^{2}= 0.67) of the variations in LAI and 96% (R

^{2}= 0.96) in NDVI improvements of 9.83% and 3.22%, respectively. Further analysis of the coefficients of estimates showed that the NDII had the most influence on variations in canopy cover with coefficient estimates of $\widehat{\xdf}$ = 0.03, −0.003 and $\widehat{\xdf}$ = 5.15, 0.97 for LST and NDII on LAI and NDVI, respectively (Figure 6 and Table 4). When analyzed for the rainy season scale (November–April), the most determinant factor of variations in the canopy cover (i.e., LAI and NDVI) were by single variable interactions. For instance, LST as a single variable accounted for 21% of variations in the LAI while interaction between LST and NDII only accounted for 24%. Further, as single variables, the LST and NDII accounted for 50% and 91% of variations in the NDVI, respectively. LST and NDII interaction also accounted for 91% variations in NDVI, the same as the NDII as a single variable (Table S8 in the supplementary data). For the dry season, the NDII as a single variable accounted for 56% variations in LAI and 90% in NDVI. As a single variable, LST accounted for 21% (R

^{2}= 0.21) variations in LAI and 56% (R

^{2}= 0.56) of the variations in NDVI (Figure 6 and Table 5).

^{2}= 0.82) of the variations in LAI and 96% (R

^{2}= 0.96) in the NDVI (Figure 6 and Table 5). Consequently, adding LST to the regression improved the outcomes by 46.43% and 6.67% in the case of LAI and NDVI, respectively (Table 5). During the annual time scale, the NDII had the most influence on the variations in the canopy cover, as indicated by the coefficients of estimates $\widehat{\xdf}$ = 0.03, −0.004 and $\widehat{\xdf}$ = 4.29, 0.93 for LST and NDII on LAI and NDVI, respectively (Figure 6 and Table 5). The interpretation of the coefficients (i.e., for the dry season), with LAI as a proxy for the canopy cover, is that the amount of change (i.e., $\widehat{\xdf}$ = 4.29) in the canopy cover (i.e., LAI) due to a 1-unit change of canopy water content (i.e., NDII) was higher than the change (i.e., $\widehat{\xdf}$

**=**0.03) in the canopy cover (i.e., LAI) due to a 1-unit change in canopy temperature (i.e., LST) (Table 5 and Tables S8–S10). For every one-unit increment in LST during the hot-dry season, the NDVI was reduced by about 0.01, while for every increase in one-unit of NDII, the NDVI increased by about 1.1 (Table S10 in the supplementary data). The model outputs and overall interpretation for the cool- and hot-dry season analyses (Tables S8–S10) were different than the annual scale results. For the hot-dry season (August–October), the LST, as a single variable, accounted for most variations (35%) in the LAI, while the NDII accounted for most variations (47%) in the NDVI. Furthermore, combination of the LST and NDII as predictors improved accounting for variations in LAI from about 35% (with either LST or NDII) to 68% (with LST and NDII combined) and NDVI from 48% (with NDII only) to 79% (with LST and NDII combined). During the cool-dry season (May–July), LST, as a single variable, had insignificant influence on variations in both LAI and NDVI. On the other hand, the NDII as a single variable accounted for 87% variations in the LAI and 94% in the variations in the NDVI. Thus, a combination of LST and NDII in the cool-dry season did not improve accounting for variations in the LAI and NDVI (Table S9).

_{Min}showing relatively stronger correlations with LAI

_{Min}and NDVI

_{Min}, there were no significant differences observed between the use of LST

_{Mean}, LST

_{Min}, and LST

_{Max}in the model.

#### 4.6. Assessment of Abrupt Change in Variables with the Pettit Test

_{Max}, and NDVI

_{Max}—to have had abrupt change in mean values between the period 2009 and 2018. Daily rainfall values also showed significant downward abrupt changes in 2012 and 2014, and an upward change was observed in 2016. The mean annual and monthly rainfall values, as well as the rest of the variables—i.e., LAI

_{Min}, LAI

_{Max}, LAI

_{Mean}, NDVI

_{Min}, NDVI

_{Mean}, NDII

_{Min}, and NDII

_{Mean}—did not show any significant abrupt changes in the time space 2009–2018 (Table 6 and Table S11a). The LST

_{Min}, LST

_{Max}, and LST

_{Mean}values seemed to have had a significant (p-value < 0.05) abrupt upward change. The NDII

_{Max}and NDVI

_{Max}seemed to have had a significant (p-value < 0.05) abrupt downward change in the mean values. At the 99% confidence level, the LST

_{Min}mean value changed from about 28.30 °C between 2009 and 2012 to 30.04 °C between 2012 and 2018. For the same time steps, the LST

_{Max}and LST

_{Mean}seemed to have changed from about 32.79 to 35.59 °C and from about 30.50 to 32.80 °C, respectively. The NDII

_{Max}mean changed from 0.05 between 2009 and 2016 to 0.01 between 2016 and 2018. Except for the daily rainfall data none of the other variables showed further change points after the initial change points.

_{Mean}in September, 2013 from LAI

_{Mean}value of 1.25 to 1.37, while the NDII

_{Max}and NDVI

_{Max}changed downwards in July from −0.03 to −0.07 and 0.52 to 0.49, respectively. There was no change in LST.

#### 4.7. Assessment of Trends in Values of Variables with Mann–Kendall Trends Test

_{Min}, LST

_{Max,}and LST

_{Mean}, indicating an upward trend, while the NDII

_{Max}and NDVI

_{Max}revealed significant downward trends with p-values of 0.0187 and 0.0022, respectively. At the alpha level = 0.05, there were significant trends in daily rainfall values (Table 7) but no significant trend in the mean monthly and annual values was observed. However, the negative sign in the S-statistic and the Kendall’s Tau for the mean monthly and annual rainfall data indicated a general downward movement. The direction of the trend is revealed by the S-statistic and the Kendall’s Tau with a positive value indicating an upward trend, and a negative value indicating a downward trend (Table 3).

_{Mean}and LAI

_{Mean}increased while NDII

_{Mean}and NDVI

_{Mean}reduced despite the insignificant (at alpha > 0.1) increase in rainfall.

## 5. Discussion

#### 5.1. Variations in Canopy Cover, Canopy Water Content and Canopy Temperature

#### 5.2. Canopy Cover Relationship with Canopy Water Content and Canopy Temperature

_{a}values increased (Figure 3 and Figure 4). This rise in the NDII and E

_{a}cannot possibly be attributed to increased root zone storage but most likely the vegetation own water storage mechanism, as espoused by Tian et al. [19] and Vinya et al. [13,83].

#### 5.3. Observed Abrupt Change Point(s) and the Trends in Variables during the 2009–2018 Period

_{Min}, LST

_{Max}, and LST

_{Mean}in August 2012. The NDVI

_{Max}and NDII

_{Max}showed abrupt downward changes in May 2010 and May 2016, respectively. The change point in LST did not correspond with the change points in NDII

_{Max}and NDVI

_{Max}(Figure 7). Therefore, from the homogeneity analysis of the 8-day values of variables the abrupt change in NDVI

_{Max}cannot be attributed to neither LST nor NDII

_{Max}as the change occurred earlier (i.e., in 2010) than that of LST and NDII in 2012 and 2016 respectively (Table 6). The 8-day NDVI

_{Max}and NDIIMax showed downward trends while the LST showed an upward trend (Table 7). Based on the CCR.LM results (Table 4), which showed that the NDII accounted for the largest variations in NDVI, the downward trend in the NDVI

_{Max}can be attributed to the downward trend in the NDII

_{Max}. Analysis of mean annual values, for the period 2009–2018, showed a significant (alpha = 0.05) upward trend in LST, increasing at an annual rate of 0.17 °C/year. The Sens’ slope showed the LAI and rainfall generally to have been in an upward trend (alpha level > 0.1), increasing at annual rates of 24.25 mm/year and 0.008/year, respectively, while the NDII and NDVI generally showed a downward trend at alpha levels 0.1 and >0.1, with annual rates of decrease at 0.001/year and 0.0001/year, respectively (Figure 8). The upward change in LST could be attributed to several factors, including but not limited to: the change in land cover (i.e., from forest to cropland or grassland), the uncertainties associated with the estimation of LST (especially during the hot-humid periods, i.e., in the rainy season), reduction in vegetation water content (as a result of a prolonged drought period or reduced rainfall), and data quality (i.e., several outliers in the data) [26,31,32,81,84]. However, in this study, it is more likely that the upward abrupt change in LST was a true change. This is because the LAI (i.e., a proxy for forest cover) showed a general upward trend, indicating that there were possibly no changes in land cover. The study site is also a conservancy with extremely limited anthropogenic activities that could result in an extensive change in land cover type. What is more probable, as could be “evidenced” by the general downward movement in NDII and NDVI (Figure 7 and Figure 8) and the daily rainfall values (i.e., as indicated by negative S and Kendall Tau values) (Table 7) is that the abrupt change and rising trend in LST may be an indication of the forest getting drier from 2009–2018 [81]. Thus, LST may be a good indicator of temporal changes in plant water status as a result of changing climatic factors in the Miombo ecosystem.

## 6. Limitations of the Study

_{a}values were not direct observations, but model-based and not validated with ground data. The E

_{a}values were based on MODIS MOD16A2, and MYD16A2, which also make use of the other indicators used (i.e., NDVI/LAI). Based on results of this study, further investigations should be made using field data to verify these satellite data-based observations.

## 7. Conclusions

## Supplementary Materials

## Author Contributions

## Funding

## Acknowledgments

## Conflicts of Interest

## References

- Guumbo, J.D.; Dumas-Johansen, M.; Muir, G.; Boerstler, F.; Xia, Z. Sustainable Management of Miombo Woodlands—Food Security, Nutrition and Wood Energy; Food and Agriculture Organization of the United Nations: Rome, Italy, 2018. [Google Scholar]
- Frost, P. The ecology of miombo woodlands. In The Miombo in Transition: Woodlands and Welfare in Africa; Centre for International Forestry Research: Borgo, Indonesia, 1996; pp. 11–57. [Google Scholar]
- Pelletier, J.; Paquette, A.; Mbindo, K.; Zimba, N.; Siampale, A.; Chendauka, B.; Roberts, J.W. Carbon sink despite large deforestation in African tropical dry forests (miombo woodlands). Environ. Res. Lett.
**2018**, 13, 1–14. [Google Scholar] [CrossRef][Green Version] - Jew, E.K.; Dougill, A.J.; Sallu, S.M.; O’Connelle, J.; Benton, T.G. Miombo woodland under threat: Consequences for tree diversity andcarbon storage and carbon storage. For. Ecol. Manag.
**2016**, 361, 144–153. [Google Scholar] [CrossRef][Green Version] - Munishi, P.K.; Mringi, S.; Shirima, D.D.; Linda, K.S. The role of the Miombo Woodlands of the Southern Highlands of Tanzania as carbon sinks. J. Ecol. Nat. Environ.
**2010**, 2, 261–269. [Google Scholar] - Sutanto, S.J.; Wenninger, J.; Coenders-Gerrits, A.M.J.; Uhlenbrook, S. Partitioning of evaporation into transpiration, soil evaporation and interception: A comparison between isotope measurements and a HYDRUS-1D model. Hydrol. Earth Syst. Sci.
**2012**, 16, 2605–2616. [Google Scholar] [CrossRef][Green Version] - Gerrits, A.M.J. The Role of Interception in the Hydrological Cycle; VSSD: Delft, The Netherlands, 2010. [Google Scholar]
- IPCC. Climate Change, Synthesis Report in Contribution of Working Groups I, II and III to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change; IPCC: Geneva, Switzerland, 2007. [Google Scholar]
- Wegehenkel, M. Modeling of vegetation dynamics in hydrological models for the assessment of the effects of climate change on evapotranspiration and groundwater recharge. Adv. Geosci.
**2009**, 21, 109–115. [Google Scholar] [CrossRef][Green Version] - Chidumayo, E.N. Species structure in Zambian Miombo woodland. J. Trop. Ecol.
**1987**, 3, 109–118. [Google Scholar] [CrossRef] - Giliba, R.A.; Boon, E.K.; Kayombo, C.J.; Musamba, E.B.; Kashindye, A.M.K.; Shayo, P.F. Species Composition, Richness and Diversity in Miombo Woodland of Bereku Forest Reserve, Tanzania. J. Biodivers.
**2011**, 2, 1–7. [Google Scholar] [CrossRef] - Gonçalves, F.M.P.; Revermann, R.; Gomes, A.L.; Aidar, M.P.M.; Finckh, M.; Juergens, N. Tree Species Diversity and Composition of MiomboWoodlands in South-Central Angola: A Chronosequence ofForest Recovery after Shifting Cultivation. Int. J. For. Res.
**2017**, 2017, 6202093. [Google Scholar] - Vinya, R.; Malhi, Y.; Brown, N.D.; Fisher, J.B.; Brodribb, T.; Aragão, L.E. Seasonal changes in plant–water relations influence patterns of leaf display in Miombo woodlands: Evidence of water conservative strategies. Tree Physiol.
**2019**, 39, 104–112. [Google Scholar] [CrossRef][Green Version] - Chidumayo, E. Climate and Phenology of Savanna Vegetation in Southern Africa. J. Veg. Sci.
**2001**, 12, 347–354. [Google Scholar] [CrossRef] - Wulder, M.A.; Franklin, S.E. Remote Sensing of Forest Environments, Introduction. In Remote Sensing of Forest Environments; Wulder, M.A., Franklin, S.E., Eds.; Springer: Boston, MA, USA, 2003. [Google Scholar]
- Lechne, A.M.; Foody, G.M.; Boyd, D.S. Applications in Remote Sensingto Forest Ecology and Management. One Earth
**2020**, 2, 405–412. [Google Scholar] [CrossRef] - Lausch, A.; Erasmi, S.; King, D.J.; Magdon, P.; Heurich, M. Understanding Forest Health with Remote Sensing—Part II—A Review of Approachesand Data Models. Remote Sens.
**2017**, 9, 129. [Google Scholar] [CrossRef][Green Version] - Wang, K.; Franklin, S.E.; Guo, X.; Cattet, M. Remote Sensing of Ecology, Biodiversity and Conservation: A Review from the Perspective of Remote Sensing Specialists. Sensors
**2010**, 10, 9647–9667. [Google Scholar] [CrossRef] [PubMed] - Tian, F.; Wigneron, J.-P.; Ciais, P.; Chave, J.; Ogee, J.; Penuelas, J.; Raebild, A.; Domec, J.-C.; Tong, X.; Brandt, M.; et al. Fensholt, Coupling of ecosystem-scale plant water storage and leaf phenology observed by satellite. Nat. Ecol. Evol.
**2018**, 2, 1428–1435. [Google Scholar] [CrossRef][Green Version] - Botta, A.; Viory, N.; Ciais, P.; Friedlingstein, P.; Monfray, P. A global prognostic scheme of leaf onset satellite data. Glob. Chang. Biol.
**2000**, 6, 709–725. [Google Scholar] [CrossRef] - Jeffers, J.N.R.; Boaler, S.B. Ecology of a miombo site. Lupa North Forest Reserve, Tanzania. I. Weather and plant growth, 1962-64. Ecology
**1966**, 54, 447–463. [Google Scholar] [CrossRef] - Woodward, F.I. Temperature and the distribution of plant species. Symp. Soc. Exp. Ecol.
**1988**, 42, 59–75. [Google Scholar] - Rouse, W.J.; Haas, R.H.; Schell, A.J.; Deering, D.W. Monitoring vegetation sytems in the great plains with ERTS. Remote Sens. Cent.
**1973**, 309–317. [Google Scholar] - Cammalleri, C.; Vogt, J. On the Role of Land Surface Temperature as Proxy of Soil Moisture Status for Drought Monitoring in Europe. Remote Sens.
**2005**, 7, 16849–16864. [Google Scholar] [CrossRef][Green Version] - Dash, P.; Gottsche, F.M.; Olesen, F.S.; Fischer, H. Land surface temperature and emissivity estimation from passive sensor data: Theory and practice-current trends. Int. J. Remote Sens.
**2002**, 23, 2563–2594. [Google Scholar] [CrossRef] - Li, Z.-L.; Tang, B.-H.; Wu, H.; Ren, H.; Yan, G.; Wan, Z.; Trigo, I.F.; Sobrino, J.A. Satellite-derived land surface temperature: Current status and perspectives. Remote Sens. Environ.
**2013**, 131, 14–37. [Google Scholar] [CrossRef][Green Version] - Avdan, U.; Jovanovska, G. Algorithm for Automated Mapping of Land Surface Temperature Using LANDSAT 8 Satellite Data. J. Sens.
**2016**, 2016, 1480307. [Google Scholar] [CrossRef][Green Version] - García-Santos, V.; Cuxart, J.; Martínez-Villagrasa, D.; Jiménez, M.A.; Simó, G. Comparison of Three Methods for Estimating Land Surface Temperature from Landsat 8-TIRS Sensor Data. Remote Sens.
**2018**, 10, 1450. [Google Scholar] [CrossRef][Green Version] - Hulley, G. MYD21A2 MODIS/Aqua Land Surface Temperature/3-Band Emissivity 8-Day L3 Global 1km SIN Grid V006; NASA EOSDIS Land Processes DAAC: Pasadena, CA, USA, 2017. [Google Scholar]
- Jones, H.G.; Rotenberg, E. Energy, Radiation and Temperature Regulation in Plants; Encyclopedia of Life Sciences; John Wiley & Sons, Ltd.: Chichester, UK, 2001; pp. 1–7. [Google Scholar]
- Sun, Z.; Wang, Q.; Batkhishig, O.; Relationship, Z.O. between Evapotranspiration and Land Surface Temperature under Energy- and Water-Limited Conditions in Dry and Cold Climates. Adv. Meteorol.
**2016**, 2016, 1–9. [Google Scholar] [CrossRef][Green Version] - Mildrexler, D.J.; Zhao, M.; Running, S.W. Aglobal comparison between station air temperatures and MODIS land surface temperatures reveals the cooling role of forests. Geophys. Res.
**2011**, 116, 1–15. [Google Scholar] [CrossRef] - Mobasheri, M.R.; Fatemi, S.B. Leaf Equivalent Water Thickness assessment using reflectance at optimum wavelengths. Theor. Exp. Plant. Physiol.
**2013**, 25, 196–202. [Google Scholar] [CrossRef][Green Version] - Fensholt, R.; Sandholt, I. Derivation of a shortwave infrared stress index from MODIS near- and shortwave infrared data in a semiarid environment. Remote Sens. Environ.
**2003**, 87, 111–121. [Google Scholar] [CrossRef] - Hardisky, M.A.; Klemas, V.; Smart, R.M. The Influence of Soil Salinity, Growth Form, and Leaf Moisture on the Spectral Radiance of Spartina alterniflora Canopies. Photogramm. Eng. Remote Sens.
**1988**, 49, 77–83. [Google Scholar] - Hunt, J.E.R.; Rock, B.N. Detection of changes in leaf water content using near and middle-infrared reflectances. Remote Sens. Environ.
**1989**, 30, 43–54. [Google Scholar] - Gao, B.C. NDWI—A normalized difference water index for remote sensing of vegetation liquid water from space. Remote Sens. Environ.
**1996**, 58, 257–266. [Google Scholar] [CrossRef] - Sriwongsitanon, N.; Gao, H.; Savenije, H.H.G.; Maekan, E.; Saengsawang, S.; Thianpopirug, S. The Normalized Difference Infrared Index (NDII) as a proxy for soil moisture storage in hydrological modelling. Hydrol. Earth Syst. Sci.
**2015**, 12, 8419–8457. [Google Scholar] [CrossRef] - Yilmaz, M.T.; Hunt, E.R., Jr.; Jackson, T.J. Remote sensing of vegetation water content from equivalent water thickness using satellite imagery. Remote Sens. Environ.
**2008**, 112, 2514–2522. [Google Scholar] [CrossRef] - Marin, C.T.; Bouten, W.; Sevink, J. Gross rainfall and its partitioning into throughfall, stemflow andevaporation of intercepted water in four forest ecosystems inwestern Amazonia. J. Hydrol.
**2000**, 237, 40–57. [Google Scholar] [CrossRef] - Myneni, R.; Knyazikhin, Y.; Park, T. MCD15A2H MODIS/Terra+Aqua Leaf Area Index/FPAR 8-day L4 Global 500m SIN Grid V006; NASA EOSDIS Land Processes DAAC: Missoula, MT, USA, 2015. [Google Scholar] [CrossRef]
- Fuller, D.O.; Prince, S.D.; Astle, W.L. The influence of canopy strata on remotely sensed observation of savannah woodlands. Int. J. Remote Sens.
**1997**, 18, 2985–3009. [Google Scholar] [CrossRef] - Cano, E.; Denux, J.-P.; Bisquert, M.; Hubert-Moy, L.; Chéret, V. Improved forest-cover mapping based on MODIStime series and landscape stratification. Int. J. Remote Sens.
**2017**, 38, 1865–1888. [Google Scholar] [CrossRef][Green Version] - Reeves, M.C.; Zhao, M.; Running, S.W. Applying improved estimates of MODIS productivity of characterize grassland vegetation dynamics. Rangel. Ecol. Manag.
**2006**, 59, 1–10. [Google Scholar] [CrossRef] - Ticehurst, C.; Guerschman, P.J.; Yun, C. The Strengths and Limitations in Using the Daily MODIS Open Water Likelihood Algorithm for Identifying Flood Events. Remote Sens.
**2014**, 6, 11791–11809. [Google Scholar] [CrossRef][Green Version] - García-Mora, T.J.; Mas, J.-F.; Hinkley, E.A. Land cover mapping applications with MODIS: A literature review. Int. J. Digit. Earth
**2012**, 5, 63–87. [Google Scholar] [CrossRef] - Vermote, E.F.; Kotchenova, S.Y.; Ray, J.P. MODIS Surface Reflectance User’s Guide. Available online: http://modis-sr.ltdri.org (accessed on 11 February 2011).
- Didan, K.; Munoz, A.B.; Solano, R.; Huete, A. MODIS Vegetation Index User’s Guide; MOD 13 Series; The University of Arizona: Tucson, AZ, USA, 2015. [Google Scholar]
- Running, S.Q. MOD16A2 MODIS/Terra Net Evapotranspiration 8-Day L4 Global 500m SIN Grid V006; NASA EOSDIS Land Processes DAAC: Missoula, MT, USA, 2017. [Google Scholar]
- Magidson, J. Correlated Component Regression: Re-thinking Regression in the Presence of Near Collinearity. In New Perspectives in Partial Least Squares and Related Methods; Abdi, H., Chin, W., Esposito Vinzi, V., Russolillo, G., Trinchera, L., Eds.; Springer: New York, NY, USA, 2013; Volume 56. [Google Scholar]
- Beilfus, R.; Risky, A. Climate for Southern African Hydro: Assessing Hydrological Risks and Consequences for the Zambezi River Basin Dams; International Rivers: Berkeley, CA, USA, 2012. [Google Scholar]
- Timberlake, J.; Chidumayo, E. Miombo Eco-Region Vision Report; Biodiversity Foundation for Africa: Bulawayo, Zimbabwe, 2001. [Google Scholar]
- Miombo Network. Available online: http://www.miombonetwork.org (accessed on 30 January 2020).
- Saha, S.; Moorthi, S.; Wu, X.; Wang, J.; Nadiga, S.; Tripp, P.; Behringer, D.; Hou, Y.; Chuang, H.; Iredell, M.; et al. The NCEP Climate Forecast System Version 2. J. Clim.
**2014**, 27, 2185–2208. [Google Scholar] [CrossRef] - Saha, S.; Moorthi, S.; Pan, H.; Wu, X.; Wang, J.; Nadiga, S.; Tripp, P.; Kistler, R.; Woollen, J.; Behringer, D.; et al. The NCEP Climate Forecast System Reanalysis. Bull. Am. Meteorol. Soc.
**2010**, 91, 1015–1057. [Google Scholar] [CrossRef] - Gray, J.; Sulla-Menashe, D.; Friedl, M.A. User Guide to Collection 6 MODIS Land Cover Dynamics (MCD12Q2) Product; NASA EOSDIS Land Processes DAAC: Missoula, MT, USA, 2019. [Google Scholar]
- Penman, H.L. Natural evaporation from open water, bare soil, and grass. Proc. R. Soc. A Math. Phys. Eng. Sci.
**1948**, 193, 120–146. [Google Scholar] - Monteith, J.L. Evaporation and environment. Symp. Soc. Exp. Biol.
**1965**, 19, 205–234. [Google Scholar] [PubMed] - Miralles, D.G.; Brutsaert, W.; Dolman, A.J.; Gash, J.H. On the use of the term ‘Evapotranspiration’. Water Resour. Res.
**2020**, 1–8. [Google Scholar] - Savenije, H.H.G. The importance of interception and why we should delete the term evapotranspiration from our vocabulary. Hydrol. Process.
**2004**, 18, 1507–1511. [Google Scholar] [CrossRef] - Tucker, C.J. Red and Photographic Infrared Linear Combination for Monitoring Vegetation. Remote Sens. Environ.
**1979**, 8, 127–150. [Google Scholar] [CrossRef][Green Version] - Nielsen, D.; Miceli-Garcia, J.J.; Lyon, D.J. Canopy cover and Leaf Area Index Relationships. Agron. J.
**2012**, 104, 1569–1573. [Google Scholar] [CrossRef][Green Version] - Knyazikhin, Y.; Glassy, J.; Privette, L.J.; Tian, Y.; Lotsch, A.; Zhang, Y.; Wang, Y.; Morisette, T.J.; Votava, P.; Myneni, B.R.; et al. MODIS Leaf Area Index (LAI) and Fraction of Photosynthetically Active RadiationAbsorbed by Vegetation (FPAR) Product (MOD15) Algorithm Theoretical Basis Document. 1999. Available online: http://eospso.gsfc.nasa.gov/atbd/modistables.html (accessed on 30 January 2020).
- Wan, Z.; Hulley, S.H.G. MYD11A2 MODIS/Aqua Land Surface Temperature/Emissivity 8-Day L3 Global 1km SIN Grid V006; NASA EOSDIS Land Processes DAAC: Missoula, MT, USA, 2015. [Google Scholar]
- Simic, A.; Fernandes, R.; Wang, S. Assessing the Impact of Leaf Area Index on Evapotranspiration and Groundwater Recharge across a Shallow Water Region for Diverse Land Cover and Soil Properties. J. Water Resour. Hydraul. Eng.
**2014**, 3, 60–73. [Google Scholar] - Yu, L.; Zeng, Y.; Su, Z.; Cai, H.; Zheng, Z. The effect of different evapotranspiration methods on portraying soil water dynamics and ET partitioning in a semi-arid environment in Northwest China. Hydrol. Earth Syst. Sci.
**2016**, 1, 975–990. [Google Scholar] [CrossRef][Green Version] - Dixon, W.J. Analysis of extreme values. Ann. Math. Stat.
**1950**, 21, 488–506. [Google Scholar] [CrossRef] - Pearson, K. Notes on regression and inheritance in the case of two parents. Proc. R. Soc. Lond.
**1895**, 58, 240–242. [Google Scholar] - Helsel, D.R.; Hirsch, R.M. Statistical Methods in Water Resources. In Hydrologic Analysis and Interpretation; U.S. Geological Survey: Washington, WA, USA, 2002; pp. 1–524. [Google Scholar]
- Cirillo, M.; Poli, A.A. On the use of the Normalised Mean Square Error in evaluating dispersion model performance. Atmos. Environ.
**1993**, 27, 2427–2434. [Google Scholar] - Pettit, A.N. A non-Parametric approach to the change-point problem. J. R. Stat. Soc. Ser. C Appl. Stat.
**1979**, 28, 126–135. [Google Scholar] [CrossRef] - Zimba, H.; Banda, K.; Anthony, C.; Wilson, P.; Peter, S.; Markus, M.; Imasiku, N. Assessment of trends in inundation extent in the Barotse Floodplain, upper Zambezi River Basin: A remote sensing-based approach. J. Hydrol. Reg. Stud.
**2018**, 15, 149–170. [Google Scholar] [CrossRef] - Kendall, M.G. Rank Correlation Measures; Charles Griffin: London, UK, 1975. [Google Scholar]
- Mann, H.B. Nonparametric test against trend. Econometrica
**1945**, 13, 245–259. [Google Scholar] [CrossRef] - Sens, P.K. Estimates of the regression coefficient based on Kendall’ Tau. Am. Stat. Assoc.
**1968**, 63, 1379–1389. [Google Scholar] - Salmi, T.; Määttä, A.; Anttila, P.; Ruoho-Airola, T.; Amnell, T. Detecting Trends of Annual Values of Atmospheric Pollutants by the Mann-Kendall Test and the Sens’ Slope Estimates: The Excel Template Application Makesens; Finnish Meteorological Institute: Helsinki, Finland, 2002. [Google Scholar]
- Fuller, D.O. Folia Phenology of Vegetation in South Central Africa and Its Relevance to Climate Change. Ph.D. Thesis, The University of Maryland, College Park, MD, USA, 1994. [Google Scholar]
- Ernst, W. Variation in the mineral content of leaves of trees in miombo woodland in South Central Africa. J. Ecol.
**1975**, 63, 801–807. [Google Scholar] [CrossRef] - Johnson, J.M.; Choinski, J.S. The spring flush in Zimbabwe’s miombo woodland. Zimb. Sci. News
**1993**, 27, 3–6. [Google Scholar] - Delogu, E.; Boulet, G.; Olioso, A.; Garrigues, S.; Brut, A.; Tallec, T.; Demarty, J.; Soudani, K.; Lagouarde, J.-P. Evaluation of the SPARSE Dual-Source Model for PredictingWater Stress and Evapotranspiration from Thermal Infrared Data over Multiple Crops and Climates. Remote Sens.
**2018**, 10, 806. [Google Scholar] [CrossRef][Green Version] - Karnieli, A.; Agam, N.; Pinker, R.T.; Anderson, M.; Imhoff, M.L.; Gutman, G.G.; Panov, N.; Goldberg, A. Use of NDVI and Land Surface Temperature for Drought Assessment: Merits and Limitations. J. Clim.
**2010**, 23, 618–633. [Google Scholar] [CrossRef] - Baldocchi, D.D.; Xu, L.; Kiang, N. How plant functional-type, weather, seasonal drought, and soil physical properties alter water and energy fluxes of an oak-grass savanah and annual grassland. Agric. For. Meteorol.
**2004**, 123, 13–39. [Google Scholar] [CrossRef][Green Version] - Vinya, R.; Malhi, Y.; Fisher, J.B.; Brown, N.; Brodribb, T.J.; Aragao, L.E. Xylem cavitation vulnerability influences tree species’ habitat preferences in Miombo woodlands. Oecologia
**2013**, 173, 711–720. [Google Scholar] [CrossRef] [PubMed] - Jaafar, S.W.M.W.; Maulud, K.N.A.; Kamarulzaman, M.A.M.; Raihan, A.; Sah, S.M.; Ahmad, A.; Saad, M.S.N.; Azmi, A.T.M.; Syukri, N.K.A.J.; Khan, W.R. The Influence of Deforestation on Land Surface Temperature—A Case Study of Perak and Kedah, Malaysia. Forests
**2020**, 11, 670. [Google Scholar] [CrossRef]

**Figure 1.**Extent of the Miombo Woodland in sub-Sahara Africa and the Luangwa sub-basin of the Zambezi River Basin showing the Miombo Woodland study location in Zambia. The studied Miombo Woodland is located within the central Zambezian Miombo Woodland, which is the largest Miombo subgroup. Map of Miombo Woodland adapted from the Miombo network [53]. Land cover map was adapted/based on the 2018 Copernicus global land cover map for Africa.

**Figure 2.**Phenology calendar at the Miombo Woodland site in Mpika, Zambia based on the data for the period 2009 to 2017. Shaded box indicates the dry season.

**Figure 3.**Illustration of (

**a**) seasonal pattern in 8-day mean values of canopy cover (i.e., LAI and NDVI), canopy water content (i.e., NDII), and canopy temperature (i.e., LST). (

**b**) Seasonal pattern in 8-day mean values of actual evaporation (E

_{a}). Shaded area for variables is the 5th and 95th percentiles. The shaded area (grey) is the dry season. The rainy and dry season seasons are clearly delineated with peak values in Ea, LAI, NDII, and NDVI observed during the rainy season while lowest values were observed in the dry season. Peak canopy temperature (i.e., LST) value(s) were observed during the dry season. E

_{a}= Actual evaporation, LAI = Leaf area index, LST = Land surface temperature, NDII = Normalised difference infrared index and NDVI = Normalised difference vegetation index.

**Figure 4.**2009–2018 January to December minimum (

**a**) and maximum (

**b**) values of E

_{a}, LAI, NDII, and NDVI. Lowest values were seen during the dry season while peak values were observed in the rainy season. Shaded areas represent the rainy season: November–December indicate start of the rainfall, January–February is the peak rainfall period, while March–April is the end of the rainfall season.

**Figure 5.**Correlation of mean values for the 8-day NDVI

_{Mean}and LAI

_{Mean}with NDII

_{Mean}(

**a**,

**b**), as well as with LST

_{Mean}(

**c**,

**d**) at the annual scale. The NDVI

_{Mean}shows stronger correlation with canopy water content (i.e., NDII) and canopy temperature (i.e., LST) compared to the LAI

_{Mean}when used as proxies for canopy cover. LST

_{Mean}shows hysteresis with NDVI

_{Mean}and LAI

_{Mean}while NDII

_{Mean}shows hysteresis with LAI

_{Mean}.

**Figure 6.**

**First row**: Plots of the CCR.LM results at annual scale (January–December) (

**a**,

**b**) and dry season (May–October) (

**c**,

**d**): LAI regression graphs at annual scale (

**a**) with NDII as only predictor variable and (

**b**) with LST and NDII used simultaneously as predictors. LAI regression graphs during the dry season (

**c**) with NDII as only predictor variable and (

**d**) with LST and NDII combined as predictor variables.

**Second row:**Plots of the CCR.LM results at annual scale (January–December) (

**a**,

**b**) and dry season (May–October) (

**c**,

**d**): NDVI regression graphs at annual scale (

**a**) with NDII as only predictor variable and (

**b**) with LST and NDII combined as predictors. NDVI regression graphs during the dry season (

**c**) with NDII as only predictor variable and (

**d**) with LST and NDII used simultaneously as predictor variables. R

^{2}= coefficient of determination, B

_{N}= unstandardized coefficient of estimate with the NDII, B

_{L}= unstandardized coefficient of estimate with the LST, i = intercept, N = number of observations.

**Figure 7.**Plots of the Pettit test for the 8-day LST

_{Min}(

**a**), LST

_{Max}(

**b**), LST

_{Mean}(

**c**), NDII

_{Max}(

**d**), NDVI

_{max}(

**e**), and mean annual rainfall (

**f**) for the period of 2009 to 2018 in the Miombo Woodland in the Luangwa Basin. Significant upward change in the LST means seemed to have occurred in August 2012. Significant abrupt changes in NDII

_{Max}and NDVI

_{Max}seem to have occurred in May at the start of the dry season.

**Figure 8.**Mann–Kendall trends test and the Sens’ slope estimates for mean annual rainfall (

**a**), mean annual LAI

_{Mean}(

**b**), mean annual NDII

_{Mean}(

**c**), mean annual NDVI

_{Mean}(

**d**), and mean annual LST

_{Mean}(

**e**) for the period of 2009 to 2018. Years with relatively high LST values seem to correspond to years with relatively low rainfall and NDII values.

**Table 1.**Characteristics of the moderate resolution imaging spectroradiometer (MODIS) terra/aqua surface reflectance data [47]. NIR = Near infrared and SWIR = Shortwave infrared.

Band No. | Visible Range | Band Centre (nm) | Wavelength Range (nm) | Resolution (m) | Key Applications |
---|---|---|---|---|---|

B01 | Red | 648 | 620–670 | 500 | Absolute land cover transformation, vegetation chlorophyll |

B02 | NIR-1 | 858 | 841–876 | 500 | Cloud amount, vegetation, land cover transformation |

B03 | Blue | 470 | 459–479 | 500 | Soil/vegetation differences |

B04 | Green | 555 | 545–565 | 500 | Green vegetation |

B05 | NIR-2 | 1240 | 1230–1250 | 500 | Leaf/canopy differences |

B06 | SWIR-1 | 1640 | 1628–1652 | 500 | Snow/cloud differences |

B07 | SWIR-2 | 2130 | 2105–2155 | 500 | Cloud properties, land properties |

Rainy Season vs. Dry Season | Dry Season Inter-Annual Variations | ||||||
---|---|---|---|---|---|---|---|

Variable | Mean Diff | DF | F | p-Value | DF | F | p-Value |

LAI_{Mean} | 2.32 | (1,454) | 596.62 | <0.001 | (9,198) | 9.41 | <0.001 |

LST_{Mean} | 2.48 | (1,438) | 20.87 | <0.001 | (9,197) | 0.59 | 0.805 |

NDII_{Mean} | 0.15 | (1,458) | 485.12 | <0.001 | (9,192) | 11.27 | <0.001 |

NDVI_{Mean} | 0.13 | (1,454) | 306.70 | <0.001 | (9,192) | 7.96 | <0.001 |

**Table 3.**Pearson (r) correlation statistics of the minimum, maximum, and mean values of LAI, LST, NDII, and NDVI at an annual scale. The LST minimum values showed relatively stronger correlation with minimum values of LAI, NDII, and NDVI compared to the maximum and mean values.

Variable | LST_{Min} | LST_{Max} | LST_{Mean} | NDVI_{Min} | NDVI_{Max} | NDVI_{Mean} | NDII_{Min} | NDII_{Max} | NDII_{Mean} | LAI_{Min} | LAI_{Max} | LAI_{Mean} |
---|---|---|---|---|---|---|---|---|---|---|---|---|

LST_{Min} | 1.00 | 0.94 | 0.97 | −0.51 | −0.38 | −0.50 | −0.42 | −0.24 | −0.37 | −0.12 | −0.02 | −0.08 |

LST_{Max} | 1.00 | 0.99 | −0.43 | −0.31 | −0.41 | −0.33 | −0.14 | −0.26 | −0.11 | 0.10 | 0.01 | |

LST_{Mean} | 1.00 | −0.43 | −0.31 | −0.41 | −0.33 | −0.15 | −0.27 | −0.11 | 0.09 | 0.01 | ||

NDVI_{Min} | 1.00 | 0.90 | 0.98 | 0.97 | 0.86 | 0.94 | 0.35 | 0.58 | 0.69 | |||

NDVI_{Max} | 1.00 | 0.94 | 0.92 | 0.93 | 0.94 | 0.36 | 0.62 | 0.72 | ||||

NDVI_{Mean} | 1.00 | 0.96 | 0.89 | 0.97 | 0.35 | 0.61 | 0.71 | |||||

NDII_{Min} | 1.00 | 0.92 | 0.97 | 0.34 | 0.65 | 0.75 | ||||||

NDII_{Max} | 1.00 | 0.96 | 0.36 | 0.70 | 0.78 | |||||||

NDII_{Mean} | 1.00 | 0.35 | 0.69 | 0.78 | ||||||||

LAI_{Min} | 1.00 | 0.20 | 0.51 | |||||||||

LAI_{Max} | 1.00 | 0.79 | ||||||||||

LAI_{Mean} | 1.00 |

**Interpretation:**Min = Minimum, Max = Maximum. Values in bold are different from 0 with a significance level alpha = 0.05.

**Table 4.**CCR.LM coefficients of estimates and goodness of fit statistics (at 95% confidence level) at annual scale with NDII and LST as predictors. Accounting for variations in LAI was improved by about 9.83% when LST and NDII were simultaneously inputted as predictors. Accounting for variations in NDVI improved by 3.22% when LST and NDII were simultaneously used as predictors.

LST | NDII | Predictors—NDII and LST | |||||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|

Model | |||||||||||||||||

Model | GFS Cv | Model | GFS CV | LST | NDII | Goodness of Fit Statistics CV | Change | ||||||||||

Variable | $\widehat{\mathit{\beta}}$ | i | R^{2} | NMSE | $\widehat{\mathit{\beta}}$ | i | R^{2} | NMSE | N | $\widehat{\mathit{\beta}}$ | $\widehat{\mathit{\beta}}$ | i | R^{2} | NMSE | N | Diff. R^{2} | (%) |

LAI | −0.001 | 1.69 | 0.13 | 0.87 | 4.69 | 1.46 | 0.61 | 0.39 | 435 | 0.03 | 5.15 | 0.61 | 0.67 | 0.33 | 435 | 0.06 | 9.83 |

NDVI | −0.01 | 0.87 | 0.17 | 0.83 | 1.02 | 0.57 | 0.93 | 0.07 | 435 | −0.003 | 0.97 | 0.67 | 0.96 | 0.04 | 435 | 0.09 | 3.22 |

**Table 5.**CCR.LM coefficients of estimates and goodness of fit statistics (at 95% confidence level) during the dry season with NDII and LST as predictors. Accounting for variations in LAI was improved by about 46% when LST and NDII were simultaneously used as predictor variable.

LST | NDII | Predictors—NDII and LST | |||||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|

Model | |||||||||||||||||

Model | GFS Cv | Model | GFS CV | LST | NDII | Goodness of Fit Statistics CV | Change | ||||||||||

Variable | $\widehat{\mathit{\beta}}$ | i | R^{2} | NMSE | $\widehat{\mathit{\beta}}$ | i | R^{2} | NMSE | N | $\widehat{\mathit{\beta}}$ | $\widehat{\mathit{\beta}}$ | i | R^{2} | NMSE | N | Diff. R^{2} | (%) |

LAI | 0.001 | 1.28 | 0.21 | 0.79 | 2.92 | 1.34 | 0.56 | 0.44 | 205 | 0.03 | 4.29 | 0.5 | 0.82 | 0.18 | 205 | 0.26 | 46.43 |

NDVI | −0.01 | 0.89 | 0.56 | 0.44 | 1.18 | 0.57 | 0.90 | 0.16 | 205 | −0.004 | 0.93 | 0.71 | 0.96 | 0.04 | 205 | 0.11 | 6.67 |

^{2}= coefficient of determination, and $\widehat{\beta}$ = unstandardized coefficient of estimate.

Variable | Period | Change Time | Year | p-Value | Alpha Level | Conf. Level | Mean from | Mean to | Test Decision |
---|---|---|---|---|---|---|---|---|---|

LAI_{Min} | 2009–2018 | 0.5463 | 0.05 | 99% | p > α, H0 accepted | ||||

LAI_{Max} | 2009–2018 | 0.6127 | 0.05 | 99% | p > α, H0 accepted | ||||

LAI_{Mean} | 2009–2018 | 0.5420 | 0.05 | 99% | p > α, H0 accepted | ||||

LST_{Min} | 2009–2018 | 4th August | 2012 | 0.0027 | 0.05 | 99% | 28.3 | 30.04 | p < α, H0 rejected |

LST_{Max} | 2009–2018 | 12th August | 2012 | <0.0001 | 0.05 | 99% | 32.79 | 35.59 | p < α, H0 rejected |

LST_{Mean} | 2009–2018 | 12th August | 2012 | <0.0001 | 0.05 | 99% | 30.5 | 32.8 | p < α, H0 rejected |

NDII_{Min} | 2009–2018 | 0.0798 | 0.05 | 99% | p > α, H0 accepted | ||||

NDII_{Max} | 2009–2018 | 16th May | 2016 | 0.0004 | 0.05 | 99% | 0.05 | 0.01 | p < α, H0 rejected |

NDII_{Mean} | 2009–2018 | 0.0942 | 0.05 | 99% | p > α, H0 accepted | ||||

NDVI_{Min} | 2009–2018 | 0.0573 | 0.05 | 99% | p > α, H0 accepted | ||||

NDVI_{Max} | 2009–2018 | 9th May | 2010 | 0.0287 | 0.05 | 99% | 0.62 | 0.56 | p < α, H0 rejected |

NDVI_{Mean} | 2009–2018 | 0.1496 | 0.05 | 99% | p > α, H0 accepted | ||||

Annual *R | 2009–2018 | 0.3162 | 0.05 | 99% | p > α, H0 accepted | ||||

Monthly *R | 2009–2018 | 0.8283 | 0.05 | 99% | p > α, H0 accepted | ||||

Daily *R | 2009–2012 | 4th April | 2012 | 0.0035 | 0.05 | 99% | 4.98 | 3.90 | p < α, H0 rejected |

Daily *R | 2012–2014 | 22nd April | 2014 | <0.0001 | 0.05 | 99% | 6.14 | 3.65 | p < α, H0 rejected |

Daily *R | 2014–2016 | 15th November | 2016 | <0.0001 | 0.05 | 99% | 2.48 | 5.07 | p < α, H0 rejected |

Variable | Mann–Kendall Statistic (S) | Kendall’s Tau | Var (S) | p-Value | Alpha | Test Interpretation | Nature of Trend |
---|---|---|---|---|---|---|---|

LST_{Min} | 8642 | 0.0819 | 10,850,159.33 | 0.0087 | 0.05 | Reject H0 | Upward |

LST_{Max} | 9305 | 0.0882 | 10,850,163.67 | 0.0047 | 0.05 | Reject H0 | Upward |

LST_{Mean} | 9709 | 0.092 | 10,850,179.67 | 0.0032 | 0.05 | Reject H0 | Upward |

NDII_{Max} | −10,094 | −0.0956 | 10,850,248.00 | 0.0022 | 0.05 | Reject H0 | Downward |

NDVI_{Max} | −7748 | −0.0734 | 10,850,248.00 | 0.0187 | 0.05 | Reject H0 | Downward |

* Rainfall 2009–2018 | −86,112 | −0.0147 | 4,830,751,280.00 | 0.2154 | 0.05 | Accept H0 | No trend |

* Rainfall 2012–2018 | −86,394 | −0.0385 | 1,143,571,362.00 | 0.0106 | 0.05 | Reject H0 | Downward |

* Rainfall 2014–2018 | 78,236 | 0.0614 | 492,345,124.00 | 0.0004 | 0.05 | Reject H0 | Upward |

* Rainfall 2016–2018 | 77,711 | 0.0609 | 493,326,659.67 | 0.0005 | 0.05 | Reject H0 | Upward |

Annual rainfall | 9.0000 | 0.2000 | 125.0000 | 0.4743 | 0.05 | Accept H0 | No trend |

Monthly rainfall | 182 | 0.0263 | 191,454.67 | 0.6791 | 0.05 | Accept H0 | No trend |

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## Share and Cite

**MDPI and ACS Style**

Zimba, H.; Coenders-Gerrits, M.; Kawawa, B.; Savenije, H.; Nyambe, I.; Winsemius, H. Variations in Canopy Cover and Its Relationship with Canopy Water and Temperature in the Miombo Woodland Based on Satellite Data. *Hydrology* **2020**, *7*, 58.
https://doi.org/10.3390/hydrology7030058

**AMA Style**

Zimba H, Coenders-Gerrits M, Kawawa B, Savenije H, Nyambe I, Winsemius H. Variations in Canopy Cover and Its Relationship with Canopy Water and Temperature in the Miombo Woodland Based on Satellite Data. *Hydrology*. 2020; 7(3):58.
https://doi.org/10.3390/hydrology7030058

**Chicago/Turabian Style**

Zimba, Henry, Miriam Coenders-Gerrits, Banda Kawawa, Hubert Savenije, Imasiku Nyambe, and Hessel Winsemius. 2020. "Variations in Canopy Cover and Its Relationship with Canopy Water and Temperature in the Miombo Woodland Based on Satellite Data" *Hydrology* 7, no. 3: 58.
https://doi.org/10.3390/hydrology7030058