- - PowerPoint PPT Presentation

!"#"$%&'()*+(,")-.,'#"*/%'$*0%'1,2-"*3-%425,1*,%* +6&-%#"*!-%2)7,*0%(',%-'()*'(*,7"*8(',"4*/,.,"1

9: ;<9*=>: ! ?@*/9;A;<*B8+:+<CD@*E;//+FG*H8F+!>I@*0 +J;*/9:>K;HL@* 0 GJ;<M;; H>=9 ?@*G<C;H*G: F8:+ ?

?! ;3G: 90 ;<9*> = *C;> C: G3 NO*P *;<A+: > <0;<9GH*: ;/ >8:F ;/@*/ > 89N ;:<*+HH+<> +/ *8<+A;:/ +9O D! ;3G: 90 ;<9*> = *C;> C: G3 NO@*> N +> / 9G9;*8<+A;: / +9O I> = = +F ;*> = *QG9;: *+<= > : 0 G9+><@*8R/ R*C;> H>C+F GH*/ 8: A;O L<G9+> <GH*QG9;: *G<! *F H+0 G9;*F ;<9;: @*8/ ! GS <: F /

6/11/18 SOIL MOISTURE-DROUGHT MONITORING 2

National Soil Moisture Network

• Daily observations from 1,500+ in situ monitoring stations

6/11/18 SOIL MOISTURE-DROUGHT MONITORING 3

nationalsoilmoisture.com

F2--"(,*3-%V"5,*>WV"5,'#"1

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

!"##"#$ %&'( *&'%+,-./0-&,12+ *&3'+&-'31 Q

3L#R&L-QH#E#Q$

So Soil Moisture Dr Drought Monitoring – Mo Models

• CPC one-layer “leaky bucket” hydrological model
• NLDAS-2
• National Water Model – WRF-Hydro
• Noah MP
• SMAP L4 surface and root zone

6/11/18 SOIL MOISTURE-DROUGHT MONITORING 5

So Soil Moisture Dr Drought Monitoring – Re Remote Sensing

• SMAP L3 Radiometer
• SMOS L3 Global Map
• SMAP/Sentinel-1 L2 Radiometer/Radar
• ESA CCI merged passive/active

6/11/18 SOIL MOISTURE-DROUGHT MONITORING 6 SMAP L3 Passive SMAP L4 Surface

Ob Objective 1: In In Situ Da Data Validation

• Relative observation error estimated

as the ratio of error variance (𝜀) to real variance (σ) of daily soil moisture (Dirmeyer et al. 2016)

• # σ

⁄ related to autocorrelation of daily soil moisture

• Surficial soil moisture exhibits higher

proportion of random observation error

• Further results suggest in-ground
• bservations exhibit less relative
• bservation error than surface-based

remote sensing observations

6/11/18 SOIL MOISTURE-DROUGHT MONITORING 7 Relative observation error of daily, summer (JJA) soil moisture from in situ monitoring networks.

𝜀 σ %

Comparison Methods

1. Determine a fixed period for each network over which to compare model/satellite datasets 2. Vertically interpolate observations to model layers, use shallowest sensor (< 10 cm) for satellites 3. Average daily soil moisture (VWC or percentiles of VWC) over all in situ stations within each model/satellite grid cell

6/11/18 SOIL MOISTURE-DROUGHT MONITORING 8

DEOS Enviroweather NOAA HMT OK Mesonet SCAN (MS & AL) SNOTel (UT & CA) SoilScape WTX Mesonet CPC

4 10 10 10 10 10 5 10

NLDAS-2

4 10 10 19 14 14 5 16

NWM

4 10 10 10 10 10 5 10

SMAP L4

3 3 3 3 3 3 3 3

SMAP L3

3 3 3 3 3 3 3 3

SMOS L3

4 7 7 7 7 7 5 7

ECV

4 10 10 19 14 14 5 16

SMAP/Sent-1

3 3 3 3 3 3 3 3

Record length (years) for in situ-model/satellite soil moisture comparison.

So Soil Moisture Da Dataset Validation – Mo Model Vari riability

6/11/18 SOIL MOISTURE-DROUGHT MONITORING 9

Standard deviation of daily summer (JJA) soil moisture from model datasets (y-axis) and in situ stations (x-axis).

Model VWC 𝜏 (cm3 cm-3) In Situ VWC 𝜏 (cm3 cm-3) NOAH 0-10 cm NOAH 10-40 cm NOAH 40-100 cm Mosaic 0-10 cm Mosaic 10-40 cm Mosaic 40-200 cm NWM 0-10 cm NWM 10-40 cm NWM 40-100 cm

NWM Variability

6/11/18 SOIL MOISTURE-DROUGHT MONITORING 10

So Soil Moisture Da Dataset Validation – Sa Satellite Variability

6/11/18 SOIL MOISTURE-DROUGHT MONITORING 11

Standard deviation of daily summer (JJA) soil moisture from satellite datasets (y-axis) and in situ stations (x-axis).

Model VWC 𝜏 (cm3 cm-3) In Situ VWC 𝜏 (cm3 cm-3) CPC SMAP L4 Surface SMAP L4 RootZone SMAP L3 ECV SMAP/Sentinel-1 SMOS L3

So Soil Moisture Da Dataset Validation – Di Diff fference (b (bias)

6/11/18 SOIL MOISTURE-DROUGHT MONITORING 12

Model/Satellite JJA VWC bias, by network and dataset.

cm3 cm-3

Mo Model Va Validation – Di Difference by by Percentile

6/11/18 SOIL MOISTURE-DROUGHT MONITORING 13

NOAH 0-10 cm NOAH 10-40 cm NOAH 40-100 cm Mosaic 0-10 cm Mosaic 10-40 cm Mosaic 40-200 cm NWM 0-10 cm NWM 10-40 cm NWM 40-100 cm

Difference (cm3 cm-3) Percentile

Sa Satellite Va Validation – Di Diff fference by by Percentile

6/11/18 SOIL MOISTURE-DROUGHT MONITORING 14

Difference (cm3 cm-3) Percentile

SMAP SMOS ECV

Drought-focused Comparison: Drought Impacts

1. NDMC Drought Impact Reporter: county-level impact reports 2. Reports collected (including impact start date) for 2008-2017 for all counties in which an in situ station resides 3. Duplicate impacts, impacts that occurred at great lags to drought conditions were removed 4. Daily VWC at all in situ stations or model/satellite grid cells over the counties averaged to achieve one, network-level dataset 5. Percentiles computed from daily VWC record, averaged to a weekly time step

6/11/18 SOIL MOISTURE-DROUGHT MONITORING 15

http://droughtreporter.unl.edu/map/

Drought-focused Comparison: Drought Impacts

1. Calculate rate at which datasets show drought conditions (< 20th percentile) corresponding to drought impacts 2. Drought “hit rates” computed for lead times ranging from 1 to 8 weeks prior to reported impact start date 3. Drought impact accuracy score is the integral of all drought hit rates, from 1 to 8 week leads; provides a measure of the reliability of datasets to show drought leading up to a drought impact

6/11/18 SOIL MOISTURE-DROUGHT MONITORING 16

http://droughtreporter.unl.edu/map/

Drought-focused Comparison: Drought Impacts

6/11/18 SOIL MOISTURE-DROUGHT MONITORING 17

• Figure shows the percent of

drought impacts reported in West Texas for which each soil moisture product showed drought conditions (< 20th percentile), as a function of the time (# weeks) prior to the reported drought impact.

Drought-focused Comparison: Drought Impacts

6/11/18 SOIL MOISTURE-DROUGHT MONITORING 18

Dataset Average Drought Score CPC 0.502 Noah 0-10 cm 0.403 Noah 10-40 cm 0.441 Noah 40-100 cm 0.474 Mosaic 0-10 cm 0.471 Mosaic 10-40 cm 0.490 Mosaic 40-200 cm 0.505 SMAP L4 Surface 0.395 SMAP L4 Root Zone 0.419 SMAP L3 0.288 SMOS L3 0.222 ECV 0.303 In Situ 0-10 cm 0.340 In Situ 10-20 cm 0.359 In Situ 20-50 cm 0.339 In Situ 50-100 cm 0.453

Top panel shows accuracy scores by dataset for drought impacts over SCAN AL network. Right table shows the average drought impact scores for all datasets, averaged over all networks.

Conclusions so far…

• Ratio of error variance (Dirmeyer et al. 2016) a solid means for assessing the fidelity
• f in situ networks/stations
• NDMC drought impacts provide tangible evidence of drought; however, working

with impact reports comes with several complexities/difficulties

• For most areas: 50-100 cm in situ sensors show drought corresponding to reported

impacts at higher rates than more surficial layers

• Models – particularly 10-100 cm layers – show drought corresponding to reported

impacts at higher rates than satellite datasets

• SMAP L4 products match best in situ daily VWC variability; SMAP L3 exhibits the

lowest difference (bias)

6/11/18 SOIL MOISTURE-DROUGHT MONITORING 19

>*("%#?-*3#@ %A1-*3B$!&=(3& /%()("#C&!*!", @ D&!*!",=EFB-$&=(3& G($$!4+#'&4!3- @ H5&4!3-B&$,$=,-) 0!I(#/%*-J(5@ B6KM5<7J;?=9:<7hO9=JI;N5JG9>)

%&'()*&'%+,-./0-&,12+) *&3'+&-'31 !"##"#$ 4E

3L#R&L-QH#E#Q$

! $)'<5L)'6-K(+5?+(>+%-'<+5>)L?,('6)% ! =$!25M()KJ.-56-,-K65>)L?,('6)%5 ! "N?,%6')%5-)5IOPQQR56-,-')%65 ! !+&+<)?L+%-H,??<'>,-')%5)S5'%-+J(,-')%5L+-.)M65

<"_,*/,"&1